RU2714100C1 - Image encoding device, image decoding device, image encoding method and image decoding method - Google Patents

Abstract

FIELD: image processing means.

SUBSTANCE: invention relates to image processing. Technical result is achieved by performing a filtering process over a decoded image, which is obtained by summing the prediction image, which is formed by using said parameter for generating prediction signals and a decoded differential image obtained by restoring a compressed differential image, based on said filtering information, wherein, when performing the filtering process, determining a class for each of the pixels constituting the decoded image, in accordance with threshold values for performance of filtration process over each of pixels containing the specified decoded image on the basis of the filter received according to certain class.

EFFECT: technical result is higher accuracy of image processing.

4 cl, 18 dwg

Description

FIELD OF THE INVENTION

The present invention relates to an image encoding apparatus and an image encoding method for compressing and transmitting an image, as well as an image decoding apparatus and an image decoding method for decoding encoded data transmitted by an image encoding apparatus to recover an image.

State of the art

Traditionally, in accordance with video encoding methods according to international standards, for example, MPEG and ITU-T H-26x, after the input video frame is divided into macroblocks, each of which is a 16 × 16 pixel block, and motion compensation prediction is performed for each macroblock, information compression is performed by performing orthogonal transformation and quantization for the prediction error signal in units of blocks.

However, the problem is that as the compression ratio becomes high, the compression efficiency decreases as a result of deterioration in quality as a prediction reference image, which is used when performing motion compensation prediction.

In order to solve this problem, in accordance with an encoding method such as MPEG-4 AVC / H.264 (see Non-Patent Reference 1), the blocking distortion that occurs in the prediction reference image when quantizing the orthogonal transform coefficients is attempted to be removed by performing blocking filtering process in the circuit.

FIG. 17 is a block diagram showing an image encoding apparatus disclosed in Non-Patent Reference 1.

In this image encoding device, when receiving an image signal that is to be encoded, the block division module 101 divides the image signal into macroblocks and outputs the image signal in units of macroblocks to the prediction module 102 as a divided image signal.

When receiving a divided image signal from the block division module 101, the prediction module 102 calculates a prediction error signal by predicting an image signal of each color component in each macroblock in a frame or between frames.

In particular, when performing prediction with motion compensation between frames, the prediction module searches for the motion vector in units of either the macroblocks themselves or each of the subblocks into which each macroblock is more accurately divided.

The prediction module then performs motion compensation prediction for the reference image signal stored in the memory 107 by using the motion vector to form a motion compensation prediction image, and determines a difference between the prediction signal showing the motion compensation prediction image and the divided image signal to calculate the prediction error signal.

The prediction module 102 also outputs the parameters for generating the prediction signals that the prediction module determines upon receipt of the prediction signal to the variable length encoding module 108.

For example, parameters for generating prediction signals include pieces of information, such as an intra prediction mode, showing how to perform spatial prediction in each frame, and a motion vector showing the amount of movement between frames.

When receiving the prediction error signal from the prediction module 102, the compression module 103 quantizes the prediction error signal to detect compressed data after performing the DCT (discrete cosine transform) process for the prediction error signal to remove the correlation of the signals from this prediction error signal.

When receiving the compressed data from the compression unit 103, the local decoding unit 104 inversely quantizes the compressed data and then performs an inverse DCT process for the compressed data inversely quantized so as to calculate a prediction error signal corresponding to the prediction error signal output from the prediction unit 102.

Upon receiving the prediction error signal from the local decoding unit 104, the adder 105 sums the prediction error signal and the prediction signal output from the prediction unit 102 to form a local decoded image.

The loop filter 106 removes the blocking distortion superimposed on the local decoded image signal showing the local decoded image generated by the adder 105 and stores the local decoded image signal with the distortion removed in the storage device 107 as a reference image signal.

When receiving the compressed data from the compression unit 103, the variable-length encoding unit 108 entropically encodes the compressed data to output a bit stream that is an encoded result.

When outputting the bitstream, variable length coding unit 108 multiplexes the parameters for generating the prediction signals output from the prediction unit 102 into a bitstream and outputs this bitstream.

According to the method disclosed in Non-Patent Reference 1, the loop filter 106 determines the smoothing intensity according to information including quantization resolution, coding mode, degree of variation of the motion vector, etc. for pixels near the border of a DCT block to provide a reduction in distortion occurring at the block boundary.

As a result, the quality of the reference image signal can be improved, and the prediction efficiency with motion compensation in subsequent coding processes can be improved.

In contrast, the problem with the method disclosed in Non-Patent Reference 1 is that the high-frequency components of the signal are lost with an increase in the compression ratio at which the signal is encoded, and therefore the entire screen is smoothed out too much and the encoded video becomes blurry.

To solve this problem, Non-Patent Reference 2 discloses a technology for using the Wiener filter as a loop filter 106 and forming this loop filter 106 in such a way that the distortion depends on the square error between the image signal to be encoded, which is the signal of the original image, and the reference image signal corresponding to this image signal is minimized.

FIG. 18 is an explanatory drawing showing a principle for improving the quality of an image reference signal using a Wiener filter in an image encoding apparatus disclosed in Non-Patent Reference 2.

In FIG. 18, the signal s corresponds to an image signal to be encoded, which is input to the block division module 101 shown in FIG. 17, and the signal s' corresponds to either a local decoded image signal output from the adder 105 shown in FIG. 17, or a local decoded image signal in which the distortion occurring at the block boundary is reduced by a loop filter 106 disclosed in non-patent reference 1.

More specifically, the signal s' is a signal in which the distortion (noise) e during encoding is superimposed on the signal s.

The Wiener filter is defined as a filter that is applied to the signal s' in such a way as to minimize this distortion (noise) e when encoding using the distortion criterion depending on the square error. Typically, filtering coefficients w can be determined by using the following equation (1) from both the signal s' auto-correlation matrix R _s's and the cross-correlation matrix R _ss' between the s and s' signals. The size of the matrices R _{s '} and R _ss' corresponds to the number of taps of a particular filter.

By applying a Wiener filter having filtration coefficients w, the signal s is “hat” whose quality is improved (“^” attached to the letter of the alphabet is referred to as “hat” because this application is an electronic patent application in Japan) is detected as signal corresponding to the reference image signal. The image encoding device disclosed in Non-Patent Reference 2 determines the filtering coefficients w in each of two or more different numbers of taps for each full frame of the image, which is the target to be encoded, and after determining a filter having the number of taps that optimizes the volume code of filtering coefficients w and distortion (e '= s "hat" -s), which is calculated after the filtering process is implemented using the distortion criterion depending on the transmission speed, additional but it divides the signal s' into many blocks having a certain size, selects whether or not to apply the Wiener filter, which has the optimal number of taps, which is determined above, to each block and transmits information on filter activation / deactivation for each block.

As a result, the additional code required to perform the Wiener filtering process can be reduced, and the image quality of the prediction can be improved.

Background Documents

Non-Patent Links

Non-Patent Reference 1. MPEG-4 AVC Standards (ISO / IEC 14496-10) /H.ITU-T 264

Non-Patent Reference 2. T.Chujoh, G.Yasuda, N.Wada, T.Watanabe, T.Yamakage, "Block-based Adaptive Loop Filter", VCEG-AI18, ITU-T Conference SG16 / Q.6, July 2008

Disclosure of Invention

Since the traditional image encoding device has the structure as described above, one Wiener filter is calculated for the entire frame, which is the goal to be encoded, information indicating whether or not to use the Wiener filtering process is applied to each of the blocks that make up every frame. However, the problem is that since an identical Wiener filter is applied to any block of each frame, a case arises when the Wiener filter is not always the optimal filter for each block, and the image quality cannot be improved sufficiently.

The present invention is implemented to solve the above problem, and therefore, the aim of the present invention is to provide an image encoding device, an image decoding device, an image encoding method and an image decoding method that can improve the accuracy of improving image quality.

In accordance with the present invention, there is provided an image encoding device in which the filtering module includes a region classification module for extracting an estimated value of each of the regions that make up the local decoded image obtained by the local decoding module in order to classify each of the regions into a class that the region belongs, according to the estimated value, and the filter compilation and processing module, for each class to which one or more regions th, from the areas that make up the local decoded image, belong, forming a filter that minimizes the error that occurs between the input image and the local decoded image in each of one or more areas belonging to the class to compensate for the distortion superimposed on one or more areas, by using a filter.

Since the filtering module in accordance with the present invention includes a region classification module for extracting an estimated value of each of the regions that make up the local decoded image obtained by the local decoding module in order to classify each of the regions into a class to which the region belongs according to the estimated quantity, and a filter compilation and processing module for, for each class that has one or more areas, from the areas that make up the local deck an odd image belong to form a filter that minimizes the error that occurs between the input image and the local decoded image in each of one or more areas belonging to the class in order to compensate for distortion superimposed on one or more areas by using a filter, the advantage of being able to increase accuracy of improving image quality.

Article I. Brief Description of the Drawings

FIG. 1 is a block diagram showing an image encoding apparatus according to Embodiment 1 of the present invention;

FIG. 2 is a block diagram showing a loop filter 6 of an image encoding apparatus according to Embodiment 1 of the present invention;

FIG. 3 is a flowchart showing a process performed by a loop filter 6 of an image encoding apparatus according to Embodiment 1 of the present invention;

FIG. 4 is an explanatory drawing showing an example of classes into which four regions (region A, region B, region C and region D) that make up the local decoded image are classified;

FIG. 5 is an explanatory drawing showing 16 blocks (K) that make up a local decoded image;

FIG. 6 is an explanatory drawing showing an example of a bit stream generated by a variable length encoding unit 8;

FIG. 7 is a block diagram showing an image decoding apparatus according to Embodiment 1 of the present invention;

FIG. 8 is a block diagram showing a loop filter 25 of an image decoding apparatus according to Embodiment 1 of the present invention;

FIG. 9 is a block diagram showing a loop filter 25 of an image decoding apparatus according to Embodiment 1 of the present invention;

FIG. 10 is a flowchart showing a process performed by a loop filter 25 of an image decoding apparatus according to Embodiment 1 of the present invention;

FIG. 11 is a flowchart showing a process performed by a loop filter 6 of an image encoding apparatus according to Embodiment 2 of the present invention;

FIG. 12 is an explanatory drawing showing an example of a selection of a Wiener filter for each of the blocks (K) that make up a local decoded image;

FIG. 13 is a flowchart showing a process performed by a loop filter 25 of an image decoding apparatus according to Embodiment 2 of the present invention;

FIG. 14 is a flowchart showing a process performed by a loop filter 6 of an image encoding apparatus according to Embodiment 3 of the present invention;

FIG. 15 is a flowchart showing a process for a first frame performed by a loop filter 6;

FIG. 16 is a flowchart showing a process for a second or subsequent frame performed by a loop filter 6;

FIG. 17 is a block diagram showing an image encoding apparatus disclosed in Non-Patent Reference 1; and

FIG. 18 is an explanatory drawing showing a principle for improving the quality of an image reference signal using a Wiener filter.

The implementation of the invention

Further, in order to explain this invention in more detail, preferred embodiments of the present invention are described with reference to the accompanying drawings.

Option 1 implementation

FIG. 1 is a block diagram showing an image encoding apparatus according to Embodiment 1 of the present invention. In FIG. 1, the block division module 1 performs the process of dividing the image signal, which is the input image and which is the target to be encoded, into macroblocks and outputting the image signal in units of macroblocks to the prediction module 2 as a divided image signal.

When receiving the divided image signal from the block division module 1, the prediction module 2 performs a prediction process for the divided image signal in the frame or between frames to generate a prediction signal.

In particular, when performing prediction with motion compensation between frames, the prediction module detects the motion vector in units of macroblocks or each of the subunits into which the macroblock is more accurately divided, both from the divided image signal and from the reference image signal showing the reference image stored in the storage device 7, to generate a prediction signal showing a prediction image from a motion vector and a reference image signal.

After generating the prediction signal, the prediction module then performs the process of calculating the prediction error signal, which is the difference between the divided image signal and the prediction signal.

In addition, when generating the prediction signal, the prediction module 2 determines the parameters for generating the prediction signals and outputs the parameters for generating the prediction signals to the variable-length encoding unit 8.

For example, parameters for generating prediction signals include pieces of information, for example, an intra-prediction mode showing how to perform spatial prediction in a frame, and a motion vector showing the amount of movement between frames.

The prediction processing module consists of a block division module 1 and a prediction module 2.

Compression unit 3 performs the process of performing the DCT (discrete cosine transform) process for the prediction error signal calculated by the prediction unit 2 to calculate DCT coefficients when quantizing the DCT coefficients to output compressed data that are DCT coefficients quantized in such a way to the local decoding unit 4 and the variable length coding unit 8. Compression module 3 constitutes a differential image compression module.

The local decoding unit 4 performs the process of performing inverse quantization of the compressed data output from the compression unit 3 and performing the inverse DCT process for the compressed data inverse quantized so as to calculate a prediction error signal corresponding to a prediction error signal output from the prediction unit 2.

The adder 5 performs the process of summing the prediction error signal calculated by the local decoding unit 4 and the prediction signal generated by the prediction unit 2 to generate a local decoded image signal showing the local decoded image.

The local decoding module consists of a local decoding unit 4 and an adder 5.

The loop filter 6 performs the filtering process to compensate for distortion superimposed on the local decoded image signal generated by the adder 5 to output the local decoded image signal, thus filtered, to the memory 7 as a reference image signal when filter information is output, which the loop filter uses when performing the filtering process, to the variable-length coding unit 8. The loop filter 6 constitutes a filter module.

The storage device 7 is a recording medium for storing a reference image signal output from the loop filter 6.

The variable length encoding unit 8 performs an entropy encoding process of compressed data output from the compression unit 3, filter information output from the loop filter 6, and parameters for generating prediction signals output from the prediction unit 2 to generate a bit stream showing these encoded results. The variable length encoding unit 8 constitutes a variable length encoding unit.

FIG. 2 is a block diagram showing a loop filter 6 of an image encoding apparatus according to Embodiment 1 of the present invention.

In FIG. 2, the frame memory 11 is a recording medium for storing only one frame of a local decoded image signal generated by the adder 5.

The area classification module 12 performs the process of extracting an estimated value of each of the areas that make up the local decoded image shown by one frame of the local decoded image signal stored in the frame memory 11 to classify each of the areas into the class to which the region belongs according to the estimated value .

The filter compilation and processing unit 13 performs the formation process, for each class to which one or more of the regions included in the regions that make up the local decoded image belong to the Wiener filter, which minimizes the error that occurs between the image signal, which is the target that should be encoded, and with a local decoded image signal in each of one or more areas that belong to the class, and using the Wiener filter to compensate for distortion, superimposed on the area.

The filter compilation and processing module 13 also performs a process of outputting the filter information by the Wiener filter to the variable length encoding unit 8.

The following explains the operation of the image encoding device.

Upon receiving the image signal, which is the target to be encoded, the block division module 1 divides the image signal into macroblocks and outputs the image signal in units of macroblocks to the prediction unit 2 as a divided image signal.

When receiving the divided image signal from the block division module 1, the prediction module 2 detects parameters for generating the prediction signals that the prediction module uses to perform the prediction process for the divided image signal in the frame or between frames. Then, the prediction module generates a prediction signal showing a prediction image using parameters for generating prediction signals.

In particular, the prediction module detects a motion vector, which is a parameter for generating prediction signals used to perform the prediction process between frames, from the divided image signal and the reference image signal stored in the storage device 7.

After obtaining the motion vector, the prediction unit 2 then generates a prediction signal by performing motion compensation prediction for the reference image signal by using the motion vector.

After generating the prediction signal showing the prediction image, the prediction unit 2 calculates a prediction error signal, which is the difference between the prediction signal and the divided image signal, and outputs the prediction error signal to the compression unit 3.

When generating the prediction signal, the prediction module 2 also determines the parameters for generating the prediction signals and outputs the parameters for generating the prediction signals to the variable-length encoding unit 8.

For example, parameters for generating prediction signals include pieces of information, for example, an intra-prediction mode showing how to perform spatial prediction in a frame, and a motion vector showing the amount of movement between frames.

When receiving the prediction error signal from the prediction module 2, the compression module 3 calculates the DCT coefficients by performing the DCT (discrete cosine transform) process for the prediction error signal and then quantizes the DCT coefficients.

The compression unit 3 then outputs the compressed data, which are DCT coefficients, quantized in this way to the local decoding unit 4 and the variable length encoding unit 8.

When receiving the compressed data from the compression unit 3, the local decoding unit 4 inversely quantizes the compressed data and then carries out the inverse DCT process for the compressed data inversely quantized so as to calculate a prediction error signal corresponding to the prediction error signal output from the prediction unit 2.

After the local decoding unit 4 calculates the prediction error signal, the adder 5 sums the prediction error signal and the prediction signal generated by the prediction unit 2 to generate a local decoded image signal showing the local decoded image.

After the adder 5 generates a local decoded image signal, the loop filter 6 performs a filtering process to compensate for distortion superimposed on the local decoded image signal, and stores the local decoded image signal thus filtered in the memory 7 as a reference image signal.

The loop filter 6 also outputs the filter information that the loop filter uses when performing the filtering process to variable length encoding unit 8.

The variable length encoding unit 8 performs an entropy encoding process of compressed data output from the compression unit 3, filter information output from the loop filter 6, and parameters for generating prediction signals output from the prediction unit 2 to generate a bit stream showing these encoded results.

At this time, although the variable-length encoding module also entropy encodes the parameters for generating the prediction signals, the image encoding device can alternatively multiplex the parameters for generating the prediction signals into a bit stream that generates an image encoding device, and output this bit stream without entropy encoding of parameters for generating prediction signals.

Next, the process performed by the loop filter 6 is specifically explained.

FIG. 3 is a flowchart showing a process performed by a loop filter 6 of an image encoding apparatus according to Embodiment 1 of the present invention.

Firstly, the frame memory device 11 of the loop filter 6 stores only one frame of the local decoded image signal generated by the adder 5.

The area classification module 12 extracts the estimated value of each of the areas that make up the local decoded image shown by one frame of the local decoded image signal stored in the frame memory 11 and classifies each of the areas into the class to which the area belongs according to the estimated value (step ST1).

For example, for each region (each block having an arbitrary size (M × M pixels)), the region classification module extracts the variance of the local decoded image signal, DCT coefficients, motion vector, quantization parameter of DCT coefficients, etc. in the region as an estimated value and performs class classification based on these pieces of information. In this case, M is an integer equal to or greater than 1.

For example, when the variance of a local decoded image signal in a region is used as an estimate if each region is classified into one of class 1-N (N is an integer equal to or greater than 1), (N-1) threshold values are prepared in advance, and the variance of the local decoded image signal is compared with each of the (N-1) threshold values (th ₁ <th ₂ <... <th _N-1 ), and the class to which the region belongs is identified.

For example, when the variance of a local decoded image signal is equal to or greater than th _N-3 and less than th _N-2 , the region is classified into class N-2. In addition, when the variance of the local decoded image signal is equal to or greater than th ₂ and less than th ₃ , the region is classified into class 3.

In this case, although an example in which (N-1) threshold values are prepared in advance is shown, these threshold values can be changed dynamically for each sequence or each frame.

For example, when using a motion vector in an area as an estimate, the area classification module calculates the average vector, which is the average of the motion vectors, or the average vector, which is the average of the motion vectors, and identifies the class to which the region belongs according to the magnitude or direction vector.

In this case, the average vector has components (components x and y), each of which is the average value of the corresponding components of the motion vectors.

On the contrary, the average vector has components (components x and y), each of which is the average value of the corresponding components of the motion vectors.

When the region classification module 12 classifies each of the regions into one of the classes 1-N, the filter compilation and processing module 13 forms, for each class, to which one or more regions included in the regions that make up the local decoded image belong, a Wiener filter, which minimizes the error that arises between the image signal, which is the target to be encoded, and the local decoded image signal in each of one or more areas that belong to Assa (Steps ST2-ST8).

For example, in case the local decoded image consists of four regions (region A, region B, region C and region D), as shown in FIG. 4, when regions A and C are classified into class 3, region B is classified into class 5, and region D is classified into class 6, the filter compilation and processing module generates a Wiener filter that minimizes the error that occurs between the image signal, which is the target that must be encoded, and a local decoded image signal in each of the areas A and C belonging to class 3.

The filter compilation and processing module additionally generates a Wiener filter, which minimizes the error that occurs between the image signal, which is the target to be encoded, and the local decoded image signal in region B, belonging to class 5, and also forms a Wiener filter, which minimizes the error arising between the image signal, which is the target to be encoded, and the local decoded image signal in region D, belonging to class 6.

For example, in the case of compiling a filter with a variable number of taps when forming a Wiener filter that minimizes error, the filter compilation and processing module 13 calculates the costs, as shown below, for each different number of taps and then determines the number of taps and filter coefficients that minimize costs .

(2)Cost = D + λ ∙ R

(2)

- where D is the sum of the quadratic errors between the image signal, which is the target to be encoded in the area to which the target filter is applied, and the filtered signal of the local decoded image, λ is constant, and R is the amount of codes that are generated in the loop filter 6.

Although in this case the costs are specified by means of equation (2), this case is only an example. For example, only the sum D of quadratic errors can be specified as costs.

In addition, another estimated value, for example, the sum of the absolute values of the error can be used instead of the sum D of square errors.

After the Wiener filter is formed for each class to which one or more regions belong, the filter compilation and processing module 13 determines whether or not each of the blocks that make up the local decoded image (for example, each of the local regions that is smaller than each of the AD regions which constitutes the local decoded image) by a block for which the filter compilation and processing module must perform the filtering process (steps ST9-ST16).

More specifically, for each of the blocks that make up the local decoded image, the filter compilation and processing unit 13 compares errors that occur between the image signal, which is the target to be encoded, and the local decoded image signal in the block between before and after the filtering process .

For example, in case the local decoded image consists of 16 blocks (K) (K = 1, 2, ..., and 16), as shown in FIG. 5, the filter compilation and processing module compares the sum of the squared errors that occur between the image signal, which is the target to be encoded, and the local decoded image signal in each block (K) between before and after the filtering process.

Block 1, block 2, block 5 and block 6, shown in FIG. 5 correspond to region A shown in FIG. 4, block 3, block 4, block 7, and block 8 shown in FIG. 5 correspond to region B shown in FIG. 4, block 9, block 10, block 13, and block 14 shown in FIG. 5 correspond to region C shown in FIG. 4, and block 11, block 12, block 15, and block 16 shown in FIG. 5 correspond to region D shown in FIG. 4.

Although the filter compilation and processing module compares the sum of the squared errors between before and after the filtering process, the filter compilation and processing module can alternatively compare either the costs (D + λ ∙ R) shown by equation (2) or the sum of the absolute error values between before and after the filtration process.

When the sum of the squared errors detected after the filtering process is less than the sum of the squared errors detected before the filtering process, the filter compilation and processing unit 13 determines that the block (K) is the block that is the target for the filtering.

In contrast, when the sum of the squared errors detected after the filtering process is equal to or greater than the sum of the squared errors detected before the filtering process, the filter compilation and processing unit determines that the block (K) is a block that is not a filtering target.

The filter compilation and processing unit 13 then calculates the costs when performing the filtering process, which leads to the cost becoming a minimum, in steps ST1 to ST16, and the costs when the filtering process of the entire frame currently being processed is not performed to determine whether to execute or there is no filtering process of the entire frame currently being processed (steps ST17-ST18).

When determining, in step ST18, to carry out the filtering process of the entire frame, the filter compilation and processing module sets the flag (frame_filter_on_off_flag) to 1 (activated) and then performs the filtering process, which leads to the fact that costs become minimum, in steps ST1-ST16, and outputs a local decoded image signal for which the filter compilation and processing module has performed the filtering process to the storage device 7 as a reference image signal (steps ST19-ST20).

For example, when a region including block (K) is region B, and the class to which region B belongs is class 5, the filter compilation and processing module performs the filtering process in block (K) by using the Wiener filter of class 5 and outputs a local decoded image signal for which the filter compilation and processing module has performed the filtering process to the storage device 7 as a reference image signal.

At this time, when determining, in steps ST1-ST16, that the costs are minimized when the process of choosing whether to perform the filtering process for each block or not is performed (during the flag (block_filter_on_off_flag) = 1 (activated)), the compilation and processing module the filter outputs the signal of the filtering to be applied to the local decoded image for the block (K), for which the filter compilation and processing module determines not to perform the filtering process, to the memory 7 as the reference image signal as it is, without executing the process filtration block (K). In contrast, when determining, in steps ST1-ST16, that the costs are minimized when the process of selecting whether to perform the filtering process for each block is not performed (during the flag (block_filter_on_off_flag) = 0 (deactivated)), the filter compilation and processing module performs the filtering process for each of all the signals of the local decoded image in the frame by using the Wiener filter of the class to which the region to which the local decoded image signal belongs is classified, and outputs a local signal th decoded image for which the preparation module and the filter processing executed the filtering process, in the storage unit 7 as a reference image signal.

On the contrary, when determining, in step ST18, the filtering process of the entire frame is not performed, the filter compilation and processing module sets the flag (frame_filter_on_off_flag) to 0 (deactivated) and outputs the signal of the filtered local decoded image to the memory 7 as the reference image signal as is (steps ST21-ST22).

In steps ST2-ST22, in the flowchart, “min_cost” is a variable for minimizing costs, “i” is an index of the number of tap [i] filter taps and a loop counter, and “j” is an index of size bl_size [j] block and cycle counter.

In addition, min_tap_idx is the index (i) of the number of filter taps at the time when costs are minimized, min_bl_size_idx is the index (j) of the block size at the time when costs are minimized, and MAX is the initial value of minimum costs (large enough value).

- tap [i] (i = 0-N1)

The sequence in which N1 (N1> = 1) of various numbers of filter taps, which are determined in advance and each of which can be selected, is stored.

- bl_size [j] (j = 0-N2)

A sequence in which N2 (N2> = 1) of different block sizes (bl_size [j] × bl_size [j] pixels), which are determined in advance and each of which can be selected, is stored.

- block_filter_on_off_flag

A flag indicating whether or not to execute the process of selecting whether to perform the filtering process for each block in the frame currently being processed.

- frame_filter_on_off_flag

A flag indicating whether or not to perform the filtering process for the frame currently being processed.

Step ST2 is a step of establishing initial values, and steps ST3-ST8 are a cycle for performing a process of selecting a number of filter taps.

In addition, step ST9 is a step of establishing initial values, and steps ST10-ST16 are a cycle for performing a block size selection process and a process of determining whether to perform a filtering process for each block having a selected block size.

In addition, steps ST17-ST18 are steps of determining whether or not to filter the entire frame currently being processed, steps ST19-ST20 are steps of performing the optimal filtering process, which is determined in steps ST1-ST16, with frame_filter_on_off_flag = 1 (activated ), and steps ST21-ST22 are the steps of setting frame_filter_on_off_flag to 0 (deactivated) and not performing the filtering process for the frame currently being processed.

After the Wiener filter is formed and then the filtering process is performed as described above, the filter compilation and processing module 13 outputs the filter information for the Wiener filter to the variable-length coding unit 8.

The filter information includes a flag (frame_filter_on_off_flag) indicating whether or not to perform the filtering process for the frame currently being processed.

When this flag is activated (indicates that the filtering process is running), information, as shown below, is included in the filter information.

(1) The number of Wiener filters (the number of classes, each of which one or more areas belong)

- The number of Wiener filters may differ for each frame.

(2) Information (index) on the number of taps of each Wiener filter

- When all filters are common in the frame, the total number of taps is turned on.

- When the number of taps is different for each filter, the number of taps of each filter is turned on.

(3) Information on the coefficients of the actually used Wiener filter (Wiener filter of each class to which one or more areas belong),

- Even if the Wiener filter is formed, the information on the Wiener filter does not turn on when the Wiener filter is not actually used.

(4) Activation / deactivation information and block size information by filters for each block

- A flag (block_filter_on_off_flag) indicating whether or not to perform the activation / deactivation operation (to perform or not the filtering process) for each block in the frame currently being processed.

- Only when block_filter_on_off_flag is activated, the block size information (index) and the activation / deactivation information on the filtering process for each block is turned on.

In this embodiment, an example in which pieces of information (1) to (4) are included in the filter information is shown. The number of Wiener filters, the number of taps of each Wiener filter and the block size for activation / deactivation can be stored both by the image encoding device and by the image decoding device as information defined together in the image encoding device and the image decoding device, instead of encoding and transmitting fragments information between them.

Furthermore, although in the above explanation of FIG. 3 is illustrated as a specific example of the process performed by the loop filter 6, steps ST9-ST16 may be omitted, and the process of not performing the activation / deactivation operation for the filtering process for each block ((4) is not included in the filter information) can be included as part the process performed by the loop filter 6.

As mentioned above, the filter information output from the filter compilation and processing unit 13 is entropy encoded by the variable length encoding unit 8 and transmitted to the image decoding apparatus.

FIG. 6 is an explanatory drawing showing an example of a bitstream generated by a variable length encoding unit 8.

FIG. 7 is a block diagram showing an image decoding apparatus according to Embodiment 1 of the present invention.

In FIG. 7, when receiving a bit stream from an image encoding device, a variable length decoding unit 21 performs a process of decoding a variable length of compressed data, filter information, and parameters for generating prediction signals that are included in the bit stream. The variable length decoding unit 21 constitutes a variable length decoding unit.

Prediction unit 22 performs the process of generating a prediction signal showing a prediction image by using parameters to generate prediction signals to which variable length decoding unit 21 has applied variable length decoding. In particular, if the motion vector is used as a parameter for generating the prediction signals, the prediction module performs the process of generating the prediction signal from the motion vector and the reference image signal stored in the storage device 26.

Prediction module 22 constitutes a prediction imaging module.

The prediction error decoding unit 23 performs the process of performing inverse quantization for the compressed data to which the decoding unit 21 applied variable length decoding, and then performing the inverse DCT process for the compressed data inverse quantized so as to calculate a prediction error signal corresponding to the prediction error signal, outputted from the prediction unit 2 shown in FIG. 1.

The adder 24 performs the process of summing the prediction error signal calculated by the prediction error decoding unit 23 and the prediction signal generated by the prediction module 22 to calculate a decoded image signal corresponding to the decoded image signal output from the adder 5 shown in FIG. 1.

The decoding module consists of a prediction error decoding module 23 and an adder 24.

The loop filter 25 performs a filtering process to compensate for distortion superimposed on the decoded image signal output from the adder 24, and then performs the process of outputting the decoded image signal thus filtered outside the image decoding device and to the memory 26 as a filtered decoded image signal. . The loop filter 25 constitutes a filter module.

The storage device 26 is a recording medium for storing a filtered decoded image signal output from the loop filter 25 as a reference image signal.

FIG. 8 is a block diagram showing a loop filter 25 of an image decoding apparatus according to Embodiment 1 of the present invention.

In FIG. 8, the frame memory 31 is a recording medium for storing only one frame of a decoded image signal output from the adder 24.

The area classification module 32 performs the process of extracting the estimated value of each of the areas that make up the decoded image shown by one frame of the decoded image signal stored in the frame memory 31 to classify each of the areas into the class to which the area belongs according to the estimated value, similarly area classification module 12 shown in FIG. 2.

The filter processing unit 33 performs the Wiener filter generation process, which applies to the class into which each area is classified by the area classification unit 32, with respect to the filter information to which the variable length decoding unit 21 has applied variable length decoding to compensate for distortion imposed on area, through the use of a Wiener filter.

Although in the example of FIG. 8, a loop filter 25, in which a frame memory 31 is set as its first stage, is shown if a closed filtering process is performed for each macroblock, a loop filter may have a structure such that the frame memory 31 located as its first stage is removed as shown in FIG. 9, and the area classification module 32 extracts an estimated value of each of the areas that make up the decoded macroblock image.

In this case, the image encoding device must perform the filtering process for each macroblock independently.

Next, the operation of the image decoding apparatus is explained.

When receiving a bit stream from an image encoding device, a variable length decoding unit 21 applies variable length decoding to compressed data, filter information, and parameters to generate prediction signals that are included in the bit stream.

When receiving parameters for generating prediction signals, the prediction module 22 generates a prediction signal from the parameters for generating prediction signals. In particular, when receiving the motion vector as a parameter for generating prediction signals, the prediction module generates a prediction signal from the motion vector and the reference image signal stored in the storage device 26.

When receiving the compressed data from the variable length decoding unit 21, the prediction error decoding unit 23 performs inverse quantization for the compressed data and then performs an inverse DCT process for the compressed data inverse quantized so as to calculate a prediction error signal corresponding to the prediction error signal output from the prediction unit 2 shown in FIG. 1.

After the prediction error decoding unit 23 calculates the prediction error signal, the adder 24 adds the prediction error signal and the prediction signal generated by the prediction unit 22 to calculate a decoded image signal corresponding to the local decoded image signal output from the adder 5 shown in FIG. . 1.

When receiving a decoded image signal from adder 24, the loop filter 25 performs a filtering process to compensate for distortion superimposed on the decoded image signal, and outputs the decoded image signal thus filtered outside the image decoding device as a filtered decoded image signal while maintaining the filtered signal the decoded image in the storage device 26 as a reference image signal.

Next, a process performed by the loop filter 25 is specifically explained.

FIG. 10 is a flowchart showing a process performed by a loop filter 25 of an image decoding apparatus according to Embodiment 1 of the present invention.

First, the frame memory 31 of the loop filter 25 stores only one frame of a decoded image signal output from the adder 24.

When the flag (frame_filter_on_off_flag) included in the filter information is activated (indicates that the filtering process is in progress) (step ST31), the area classification module 32 extracts an estimated value of each of the areas that make up the decoded image shown by one frame of the decoded image signal stored in the storage device 31 frames, and classifies each of the areas into the class to which the area belongs, according to an estimated value, similar to the area classification module 12 shown ohm in FIG. 2 (step ST32).

When receiving the filter information from the variable length decoding unit 21, the filter processing unit 33 generates a Wiener filter that is applied to the class to which each of the areas classified by the area classification module 32 belongs with respect to the filter information (step ST33).

For example, when the number of Wiener filters (the number of classes, each of which one or more areas belong) is expressed as N, the number of taps of each Wiener filter is expressed as L × L, and the coefficient values of each Wiener filter are expressed as w _i11 , w _i12 , ..., w _i1L , ..., w _iL1 , w _iL2 , ..., w _iLL , N filters W _i Wiener (i = 1, 2, ..., N) is shown as follows.

After forming the N filters W _i Wiener processing unit 33, the filter compensates for distortion imposed decoded image signal by one frame through the use of the Wiener filter and outputs the signal decoded image distortion compensation outside the picture decoding device and the memory device 26 as a signal filtered decoded image (step ST34).

The signal “hat” of the filtered decoded image is expressed by the following equation (4).

The matrix S is a group of reference signals in L × L pixels, including a decoded image signal s, which is the filtering target, and id (s) is the class number (filter number), which is determined by the area classification module 32 and to which the region including the signal s, belongs.

When performing the above filtering process, the filter processing unit 33 accesses the flag (block_filter_on_off_flag) included in the filter information, and when the flag (block_filter_on_off_flag) is set to 1 (activated), accesses the block size information included in the filter information, and then identifies a plurality of blocks (K) that make up the decoded image, and then performs a filtering process on the information included in the filter information and indicating whether or not to perform the filtering process for each block (K).

More specifically, when the flag (block_filter_on_off_flag) is set to 1 (activated), the filter processing unit 33 performs a filtering process for the decoded image signal in block (K) in which the filtering unit is to perform the filtering process from the blocks that make up the decoded image by use the Wiener filter of the class to which the region including the block (K) belongs, when outputting the signal of the filtering to be decoded, the decoded image in the block (K) in which the filtering module should not filtering process, outside the picture decoding device and the memory device 26 as filtered decoded image signal as it is.

In contrast, when the flag (block_filter_on_off_flag) is set to 0 (deactivated), the filter processing module performs a filtering process for each of all decoded image signals in the frame currently being processed by using a filter corresponding to the class into which each area is classified by the module 32 area classifications.

When the flag (frame_filter_on_off_flag) included in the filter information is deactivated (filtering process is not performed) (step ST31), the filter processing unit 33 does not perform the filtering process for the frame currently being processed and outputs each decoded image signal output from the adder 24 , outside the image decoding device and to the storage device 26 as a filtered decoded image signal as is (step ST35).

As can be seen from the above description, in the image encoding apparatus according to this embodiment 1, the loop filter 6 includes a region classification module 12 for extracting an estimated value of each of the regions that make up the local decoded image shown by the local decoded image signal, output by adder 5 to classify each of the regions into the class to which the region belongs, according to the estimated value, and the module 13 compiling and processing the filter for, for each class to which one or more areas, from the areas that make up the local decoded image, belong, forming a Wiener filter that minimizes the sum of the squared errors that occur between the image signal, which is the goal that should be encoded, and with a local decoded image in each of one or more areas belonging to the class, to compensate for the distortion superimposed on one or more areas by means of olzovaniya Wiener filter. Therefore, the image encoding device implements the filtering process according to local image properties, thereby allowing to increase the accuracy of improving image quality.

Furthermore, in the image decoding apparatus according to this embodiment 1, the loop filter 25 includes an area classification module 32 for extracting an estimated value of each of the areas that make up the decoded image shown by the decoded image signal output by the adder 24 so that classify each of the regions into the class to which the region belongs, according to the estimated value, and the filter processing module 33 to access the filter information To which decoding variable length applied decoding variable length unit 21 to generate Wiener filter, which is applied to a class to which each area classified by the module 32 classifying areas belongs, and to compensate for distortion imposed on the region, by using a Wiener filter. Therefore, the image decoding apparatus implements the filtering process according to the local properties of the image, thereby allowing to increase the accuracy of improving image quality.

Option 2 implementation

In the above embodiment 1, a loop filter is shown in which the filter compilation and processing unit 13 generates a Wiener filter for each class to which one or more areas belong, and performs a filtering process for each of the blocks (K) that make up the local decoded image by use the Wiener filter of the class to which the region including the block (K) belongs. Alternatively, for each of the blocks, the loop filter can select a Wiener filter that minimizes the sum of square errors that occur between the image signal, which is the target to be encoded, and the local decoded image signal in block (K), from Wiener filters, which the loop filter forms for each class to which one or more areas belong, and can compensate for the distortion superimposed on the block (K) by using the Wiener filter selected way.

Specifically, the loop filter of this embodiment operates as follows.

FIG. 11 is a flowchart showing a process performed by a loop filter 6 of an image encoding apparatus according to Embodiment 2 of the present invention.

The filter compilation and processing module 13 generates a Wiener filter for each class to which one or more areas belong, similar to the filter compilation and processing module in accordance with the above embodiment 1 (steps ST2-ST8).

In accordance with this embodiment 2, the filter compilation and processing module does not use a flag (block_filter_on_off_flag) indicating whether or not to perform the filtering process for each block in the frame currently being processed, but using the flag ( block_filter_selection_flag), indicating whether or not to select the filter that should be used for each block in the frame currently being processed. In addition, the flag (block_filter_selection_flag) is initially deactivated in step ST40 and is activated only when step ST46 is performed.

As mentioned below, only when the flag (block_filter_selection_flag) is activated, the block size and filter selection information for each block are included in the filter information.

After the Wiener filter is formed for each class to which one or more areas belong, the filter compilation and processing module 13 selects the optimal process (for example, a process that minimizes the sum of the square errors that occur between the image signal, which is the target to be encoded, and a local decoded image signal in block (K)), from the process of filtering for each of the blocks (K) that make up the local decoded image, by ora Wiener filter of the Wiener filter, which drawing unit and a filter processing for each class generates, which one or more regions belong and the process of failure of the filtering process for each of the blocks (steps ST9 and ST41-ST47).

More specifically, in the case of the formation of four Wiener filters W ₁ , W ₂ , W ₃ and W ₄ and performing the filtering process using each of these four Wiener filters, the filter compilation and processing module selects a Wiener filter W ₃ that minimizes the sum of E squared errors for block (K), if the sum of E squared errors in block (K) has the following inequality between the four filters.

E _W3 <E _W2 <E _W4 <E _W0 <E _W1

- where E _W0 shows the sum of E squared errors at the moment when the filtering process is not performed at all.

FIG. 12 is an explanatory drawing showing an example of selecting a Wiener filter for each of the blocks (K) that make up a local decoded image. For example, a Wiener filter W ₂ is selected for block (1), and a Wiener filter W ₃ is selected for block (2).

When determining to perform the filtering process for the frame currently being processed by using the selected Wiener filters, the filter compilation and processing module 13 sets the flag (frame_filter_on_off_flag) to 1 (activated) and performs a filtering process that minimizes costs in steps ST1-ST9 and ST40 -ST47 and outputs the local decoded image signal thus filtered to the memory 7 as a reference image signal (steps ST17-ST20).

On the contrary, when determining not to filter the entire frame that is currently being processed (steps ST17-ST18), the filter compilation and processing module sets the flag (frame_filter_on_off_flag) to zero (deactivated) and outputs the signal to be filtered from the local decoded image to the memory 7 as an image reference signal (steps ST21-ST22).

After the Wiener filters are formed and then the filtering process is performed in the aforementioned manner, the filter compilation and processing module 13 outputs the filter information for the Wiener filters to the variable length coding unit 8.

A flag (frame_filter_on_off_flag) indicating whether or not to perform the filtering process in the frame currently being processed is included in the filter information.

When this flag is activated (indicates that the filtering process is running), information, as shown below, is included in the filter information.

(1) The number of Wiener filters (the number of classes, each of which one or more areas belong)

- The number of Wiener filters may differ for each frame.

(2) Information (index) on the number of taps of each Wiener filter

- When all filters are common in the frame, the total number of taps is turned on.

- When the number of taps is different for each filter, the number of taps of each filter is turned on.

(3) Information on the coefficients of the actually used Wiener filter (Wiener filter of each class to which one or more areas belong),

- Even if the Wiener filter is formed, the information on the Wiener filter does not turn on when the Wiener filter is not actually used.

(4) Filter selection information for each block and block size information

- A flag (block_filter_selection_flag) showing whether or not to select a filter for each block in units of frames.

- Only when block_filter_on_off_flag is activated, block size information (index) and selection information for each block is included.

In this embodiment, an example in which pieces of information (1) to (4) are included in the filter information is shown. The number of Wiener filters, the number of taps of each Wiener filter, and the block size can be stored both by the image encoding device and by the image decoding device as information defined together in the image encoding device and the image decoding device, instead of encoding and transmitting pieces of information between them.

The loop filter 25 in the image decoding apparatus performs the following process.

FIG. 13 is a flowchart showing a process performed by a loop filter 25 of an image decoding apparatus according to Embodiment 2 of the present invention.

First, the frame memory 31 of the loop filter 25 stores only one frame of a decoded image signal output from the adder 24.

When the flag (frame_filter_on_off_flag) included in the filter information is activated (indicates that the filtering process is running) (step ST31), and when the flag (block_filter_selection_flag) included in the filter information is deactivated (step ST51), the area classification module 32 retrieves the estimated value each of the regions that make up the decoded image shown by one frame of the decoded image signal stored in the frame memory 31 and classifies each of the regions into a class to which the region belongs um, according to the estimated value (step ST32), similar modulo classifying regions according to the above embodiment 1. FIG.

In contrast, when the flag (frame_filter_on_off_flag) included in the filter information is activated (indicates that the filtering process is running) (step ST31), and when the flag (block_filter_selection_flag) included in the filter information is activated (step ST51), the area classification module refers to information on the size of each block, which is a unit for selection, and filter selection information for each block of pieces of information included in the filter information, and performs class classification for each block (step ST52).

After the region classification module 32 classifies each region (each block) into a class to which the region belongs, the filter processing module 33 accesses the filter information output from the variable length decoding section 21 and generates a Wiener filter that is applied to the class to which each region (each block) classified by the region classification module 32 belongs to (step ST33), similarly to the filter processing module in accordance with the above embodiment 1.

After generating a Wiener filter that applies to each class when (block_filter_selection_flag) is deactivated, the filter processing module 33 performs a filtering process for each of all decoded image signals in the frame currently being processed by using the generated Wiener filters and outputs each decoded image signal thus filtered outside of the image decoding device and into the memory 26 as a filtered decoded signal and siderations (step ST53), as in the case where the flag (block_filter_on_off_flag) has been deactivated in the above embodiment 1.

In contrast, when (block_filter_selection_flag) is activated, the filter processing unit 33 compensates for the distortion applied to the decoded image signal in each block by using the Wiener filter that is selected for the block after generating the Wiener filter that is applied to each class and outputs the decoded image signal, thus filtered, outside of the image decoding apparatus and into the storage device 26 as a filtered decoded image signal (step ST53).

The signal “hat” of the filtered decoded image at this time is expressed by the following equation (5).

The matrix S is a group of reference signals in L × L pixels, including a decoded image signal s, which is a filtering target.

- id_2 (bl) is the filter selection information in bl, in which the signal s of the decoded image is included, i.e. the class number (filter number) of the bl block.

- id_2 (bl) = 0 shows a block for which the filtering process is not performed at all. Therefore, the filtering process is not performed at all in the block.

As can be seen from the above description, since the image encoding apparatus according to this embodiment 2 has such a structure that for each of the blocks (K) that make up the decoded image, the loop filter selects a Wiener filter that minimizes the sum of the squared errors that occur between an image signal, which is the target to be encoded, and a decoded image signal in block (K), from Wiener filters, which the loop filter generates for each of the class to which one or more regions belong, and compensates for the distortion applied to the block (K) by using the Wiener filter selected in this way, there is the advantage of further improving the accuracy of improving image quality compared to the above embodiment 1.

Option 3 implementation

In the above embodiment 2, a method of selecting from a process of performing a filtering process for each of the blocks (K) that make up the decoded image by using one of the Wiener filters that are generated for each class that has one or more areas in the frame, currently time processed, belong, and the process of not fulfilling the filtering process for each block, a process that minimizes the sum of the quadratic errors that occur between the image signal, which is The goal to be encoded and the local decoded picture signal in block (K). Alternatively, from the process of preparing one or more Wiener filters in advance and using one of the one or more Wiener filters that are prepared in advance, the process of using one of the Wiener filters that are generated for each class that has one or more areas in the frame is currently time processed, belong, and the process of not fulfilling the filtering process for each block, the loop filter can choose a process that minimizes the sum of the quadratic errors that occur between the signal The images, which is the aim of which is to be coded, and the local decoded picture signal in the block (K).

FIG. 14 is a flowchart showing a process performed by a loop filter 6 of an image encoding apparatus according to Embodiment 3 of the present invention.

Since this embodiment 3 provides a wider selection of Wiener filters compared to the selection in the above embodiment 2, the likelihood that an optimal Wiener filter is selected is increased compared to the above embodiment 2.

Since the method for selecting a Wiener filter is identical to the method shown in the above embodiment 2, an explanation of the method is omitted hereinafter.

Since the process performed by the image decoding apparatus is identical to the process according to the above embodiment 2, an explanation of the process will be omitted hereinafter.

Option 4 implementation

In the above embodiment 2, a method of selecting from a process of performing a filtering process for each of the blocks (K) that make up the decoded image by using one of the Wiener filters that are generated for each class that has one or more areas in the frame, currently time processed, belong, and the process of not fulfilling the filtering process for each block, a process that minimizes the sum of the quadratic errors that occur between the image signal, which is The goal to be encoded and the local decoded picture signal in block (K). Alternatively, from the process of using one of the Wiener filters that are generated for each class to which one or more areas in the frame currently being processed belong to the process of using one of the Wiener filters that are used for an already encoded frame, and the process of non-execution filtering process for each block, the loop filter can select a process that minimizes the sum of the quadratic errors that occur between the image signal, which is the goal that should be encoded, and a local decoded picture signal in block (K).

FIG. 15 is a flowchart showing a process for a first frame that is performed by a loop filter 6 of an image encoding device, and is identical to the flowchart shown in FIG. 11 in the above embodiment 2 implementation.

FIG. 16 is a flowchart showing a process for a second frame and subsequent frames that is performed by a loop filter 6.

As a reference method for accessing a Wiener filter that is used for an already encoded frame, for example, reference methods, as shown below, may be provided.

Method (1) of accessing the Wiener filter, which is used for the block at the position shown by the characteristic motion vector, which is calculated in the block, which is the target for filtering.

Method (2) of accessing the Wiener filter, which is used for a block located in a frame that is closest in time to a block that is a filter target and is located at a position identical to the position of the target block.

Method (3) of accessing the Wiener filter, which is used for a block having the greatest cross-correlation between blocks in an already encoded frame.

In the case of using method (3), an identical block search process should be performed by an image encoding device and an image decoding device.

Since this embodiment 4 provides a wider selection of Wiener filters compared to the selection in the above embodiment 2, the likelihood that an optimal Wiener filter is selected is increased compared to the above embodiment 2.

Since the method for selecting a Wiener filter is identical to the method shown in the above embodiment 2, an explanation of the method is omitted hereinafter.

Since the process performed by the image decoding apparatus is identical to the process according to the above embodiment 2, an explanation of the process will be omitted hereinafter.

Industrial applicability

An image encoding device, an image decoding device, an image encoding method, and an image decoding method in accordance with the present invention can improve the accuracy of improving image formation quality. An image encoding apparatus and an image encoding method are suitable for use as an image encoding apparatus and the like. and a method for encoding images and the like. for compression encoding and image transmission, respectively, both the image decoding apparatus and the image decoding method are suitable for use as an image decoding apparatus and the like. and a method for decoding images and the like. for decoding encoded data transmitted by the image encoding device to recover the image, respectively.

Claims (36)

1. An image decoding device comprising: - variable length decoding means that performs a variable length decoding process on an encoded bit stream to obtain a parameter for generating prediction signals, a compressed differential image, and filtering information; and - filtering means that performs the filtering process on the decoded image, which is obtained by summing the prediction image generated by using said parameter to generate the prediction signals and the decoded differential image obtained by reconstructing said compressed differential image based on said filtering information; wherein the variable-length decoding means performs a variable-length decoding process on information associated with a plurality of blocks that comprise said decoded image, said information is identification information that indicates for each of said blocks containing a decoded image whether to perform the filtering process or not, wherein said filtering means determines the class of each of the pixels that make up the specified decoded image in accordance with threshold values, to perform the filtering process on each of the pixels that make up the specified decoded image based on a filter obtained in accordance with a specific class, and not performing the filtering process on the block with respect to which the identification information indicates that the filtering process should not be performed, and however, the module for determining the specified class is different from the specified block, the corresponding information that indicates whether to perform the filtering process or not. 2. A method of decoding images containing: - a step of performing a variable length decoding process over an encoded bit stream to obtain a parameter for generating prediction signals, a compressed differential image, and filtering information; and - a step of performing the filtering process on the decoded image, which is obtained by summing the prediction image, which is generated by using the mentioned parameter to generate prediction signals and the decoded differential image obtained by restoring the compressed differential image, based on the specified filtering information; at the same time, at the stage of obtaining the specified filtering parameter, information related to a plurality of blocks that contain the specified decoded image is obtained, wherein said information is identification information that indicates for each of the blocks containing said decoded image whether to perform the filtering process or not; moreover, when performing the filtering process, determine the class for each of the pixels that make up the specified decoded image, in accordance with threshold values for performing the filtering process on each of the pixels containing the specified decoded image based on the filter obtained in accordance with a specific class, and not performing the filtering process on the block, regarding which identification information indicates that the filtering process should not be performed, the module for determining the specified class is different from the block corresponding to the information that indicates whether to perform the filtering process or not. 3. An image encoding device comprising: - a prediction means for performing a prediction process on each of the blocks of the input image to obtain a plurality of prediction images and parameters for generating prediction signals for said blocks; - compression means for performing the compression process on the differential image between the specified prediction image and the input image to obtain a compressed differential image; a filtering means for performing a filtering process on a decoded image obtained by summing a prediction image that is generated using a parameter to generate a prediction signal and a reconstructed differential image that is generated by performing the recovery process on the compressed differential image; and - variable length encoding means for performing a variable length encoding process on said parameter for generating a prediction signal, said compressed differential image and filtering information, for use in the filtering process; wherein the variable-length encoding means performs a variable-length decoding process on information associated with a plurality of blocks that comprise said decoded image, said information is identification information that indicates for each of said blocks containing a decoded image whether to perform the filtering process or not, wherein said filtering means determines the class of each of the pixels that make up the specified decoded image in accordance with threshold values and performs a filtering process on each of the pixels that make up the specified decoded image based on a filter obtained in accordance with a specific class, and failing to perform the filtering process on the block regarding which identification information indicates that the filtering process should not be performed, and the module for determining the specified class is different from the block corresponding to the information that indicates whether to perform the filtering process or not. 4. A method of encoding images containing: - a step for performing a prediction process on each of the blocks of the input image to obtain a plurality of prediction images and parameters for generating prediction signals for said blocks; a step of performing the compression process on the differential image between the specified prediction image and the specified input image to obtain a compressed differential image; - a step of performing the filtering process on the decoded image obtained by summing the prediction image, which is generated using the specified parameter to generate the prediction signal and the reconstructed differential image, which is generated by performing the recovery process on the compressed differential image; and - a stage of the variable-length coding process above the specified parameter for generating a prediction signal, the specified compressed differential image and filtering information for use in the filtering process; at the same time, at the stage of the variable-length encoding process, the variable-length encoding process is performed on the information associated with a plurality of blocks that contain the specified decoded image, said information is identification information that indicates for each of said blocks containing a decoded image whether to perform the filtering process or not, at the same time, at the stage of the filtering process, the class of each of the pixels that make up the specified decoded image is determined in accordance with threshold values and the filtering process is performed on each of the pixels that make up the specified decoded image based on a filter obtained in accordance with a specific class, and failing to perform the filtering process on the block regarding which identification information indicates that the filtering process should not be performed, and however, the module for determining the specified class is different from the specified block, the corresponding information that indicates whether to perform the filtering process or not.

Images